EP3271875A1

EP3271875A1 - System and method for predicting solar power generation

Info

Publication number: EP3271875A1
Application number: EP16710248.2A
Authority: EP
Inventors: Alexandros CHARALAMBIDES; Rogiros TAPAKIS
Original assignee: Cyprus University Of Technology
Current assignee: Cyprus University Of Technology
Priority date: 2015-03-17
Filing date: 2016-03-17
Publication date: 2018-01-24
Also published as: GB201504505D0; WO2016146788A1

Abstract

A system and a method for predicting the power output from a network of solar power generators, which includes measuring the power output of the network; storing the measured power output of the network; and predicting an expected power output of the network. The prediction of subsequent power output includes generating a power attenuation map; calculating a dynamic flow map for power attenuation based on a plurality of sequentially generated power attenuation maps; and predicting the expected power output of the network based on the measured power output and the dynamic flow map for power attenuation.

Description

System and method for predicting solar power generation

Technical Field

The present invention relates to the field of solar power generators and in particular, but not exclusively, to a system for predicting the power output from a network of solar power generators. The invention further relates to a method for predicting the power output from a network of solar power generators.

Background Art

Solar technologies have recently made significant progress in the field of electricity generation, however, solar power generators have not yet achieved high penetration in the energy sector compared to the existing fossil fuel power plants and hydroelectric power plants, mainly due to their dependency on solar irradiance. A reduction of solar irradiance over a solar power generator due to moving clouds results in an unexpected reduction in power output. However, at the same time, the demand for electricity is not changed, causing a sudden disparity between the electricity production and demand. This disparity, along with the voltage drop due to the sudden decrease in power output, sets the stability of the grid at risk. To achieve more efficient grid management, operators of large scale networks of solar power generators in some countries or regions are required to provide both day-ahead and intra-day expectations of the power output of the network. These expected power output estimations may be based on, for example, historical data relating to the location and time of year, or may be a calculated value based on a computer model of the system.

Deviation from the provided forecast leads to penalty impositions, resulting in reduced profit for the operator, while an accurate forecast may result in additional bonuses. This requirement motivates the operators of solar power generation networks to prioritize a stable network performance over maximum output, in order to accurately match the expected electricity production of their plant for a specified time horizon. Operators may restrict the overall power output of the network to match an expected level of output, thereby reducing the overall efficiency of the network and failing to maximise the power generating potential of the system. Some operators of networks of solar power generators use meteorological data to improve the calculation of expected power output for a given time horizon. Such data may be obtained from meteorological satellites and may require substantial additional processing by local and regional meteorological offices.

Operation of these networks is dependent upon extrinsic data provided by a third party. There may be delays in obtaining meteorological data, particularly for very short time horizons e.g. intra-day predictions. It may be expensive to obtain the data required for a certain time horizon. Additionally, deficiencies in the meteorological data may lead to inaccurate predictions of power output, particularly for very short-term intra-day forecasts. Obtained data may be inaccurate or incomplete, or may be of a low

resolution.

Deficiencies in data obtained from a third party must be accounted for in the expected power output. To avoid penalty impositions, the operator may provide a conservative estimate for the system output and restrict power generation to match the expected output value. Deficiencies in extrinsic data can therefore reduce the overall efficiency of the system.

Alternatively, an operator may account for a forecast with a lower accuracy by using electricity generated by a base load power plant, such as coal, gas, hydro or nuclear power plants, which can generate dependable power to consistently meet demand. A base load power plant can generate electricity to compensate for any shortfall in the power output of the network of solar power generators. However, it is impossible to control the stability of the power grid continuously during the day by adjustment of a base load power plant, as these power plants typically exhibit a slow response to changing demand. Therefore, base load power plants are generally configured by their operators to produce excess electricity, in order to compensate for potential

fluctuations in the power output of the network of solar power generators. This may result in reduced system efficiency and also cause undesirable C0₂ emissions arising from the excess production of energy by fossil fuel power plants.

It is an aim of the present invention to provide an improved prediction of the power output of a network of solar power generators over a very short-term time horizon. Summary of Invention

According to an aspect of the invention, there is provided a system for predicting the power output of a network of solar power generators according to claim 1. According to a further aspect of the invention, there is provided a method for predicting the power output of a network of solar power generators according to claim 10.

Additional features are provided as set out in the dependent claims. The system and method of the present invention provide an accurate forecast for the power output of a network of solar power generators using data which are intrinsic to the network. The forecast provided by the present invention is particularly accurate over a very short-term time horizon and therefore avoids the need for compensation using base load power plants, which leads to a reduction in C0₂ omissions across the electricity power grid.

Brief Description of the Drawings

Embodiments of the invention will now be described, by way of example, with reference to the accompanying drawings, in which:

Figure 1 is a schematic diagram of a system according to a first embodiment;

Figure 2 is a schematic diagram of a processing server of the first embodiment;

Figure 3 is a schematic representation of a normalisation process; and

Figure 4 is a flow diagram of a method according to a second embodiment. Detailed Description

With reference to Figure 1, a system 1 according to the first embodiment is shown, which is configured to predict the power output from a network of solar power generators 100. The system 1 comprises monitoring means 10 connected to the network of solar power generators 100, a database 20 for storing data received from the monitors 10 and a processing server 30 for processing data received from the monitors 10 and predicting an expected power output of the network of solar power generators 100.

The present invention can provide an accurate forecast for the power output of a network of solar power generators 100 based only on the measured power output of the network. The invention therefore avoids the need for additional exogenous data, such as meteorological data, which may be difficult to obtain and may include deficiencies. Furthermore, an operator of a network of solar power generators 100 is able to supply the maximum amount of energy capable, in accordance with an accurate prediction. The present invention thereby improves network efficiency in compliance with national legislation, increasing confidence in the solar energy sector. In this way, the

profitability of a network of solar power generators 100 may also be improved by avoiding penalty impositions for inaccuracy.

The plurality of solar power generators 100 forming the network each comprise one or more photovoltaic (PV) panels. Solar power generators may include commercial PV parks comprising a large number of PV panels, including concentrating PVs, as well as small scale PV installations, such as residential installations. Solar power generators may be on trackers or stationary, ground mounted, roof mounted, mounted on shading devices or building integrated. In some embodiments, solar power generators 100 may also include solar thermal power plants.

The system of the present invention is adapted to accumulate data from a geographic region with a number of solar power generators having different sizes and

configurations. A greater density of power generation sites within the geographical region provides a more accurate forecast for power generation across the region. New solar power generators which are added to the network may be integrated with the system to further improve accuracy over time.

The monitoring means comprises a plurality of monitors 10, each of which is connected to one or more solar power generators 100, and is configured to measure the power output of each generator in the network. The measured power output is sent to the database 20 and to the processing server 30 by the plurality of monitors 10. The monitors 10 may be configured to measure the power output in a continuous manner, or otherwise may measure the power output at a predetermined sample rate. The time between measurements may be in the range of o to 5 minutes and, in some

embodiments, may be up to 15 minutes.

The system 1 may include a monitor connected to each PV panel in a PV park, or integrated with the PV panels themselves. Alternatively, the power output for a plurality of local PV panels may be measured by a single monitor which is connected thereto, e.g. a monitor on an inverter connected to multiple PV panels. In some embodiments, the monitoring means 10 comprises a single monitor, configured to measure the power output of each solar power generator by repeatedly scanning the network. A monitor 10 which is connected to one or more solar power generators loo may detect the addition of a further solar power generator and measure the power output of the newly added generator. The newly added generator may include a new commercial PV park comprising a large number of PV panels, or may be a small scale PV installation, such as a residential installation. In some cases, the newly added generator may be one or more additional PV panels which are added to an existing installation, e.g. an increase in the number of PV panels connected to a single inverter. The system l of the present invention requires only a measurement of the power output and therefore solar power generators of any size and configuration can be integrated into the system l. The database 20 is configured to receive and store the power output for the solar power generators 100 in the network, as measured by the monitoring means 10. The database 20 is configured to accumulate measurements of power output provided by the one or more monitors 10 over time. The system 1 may be configured such that the database 20 receives a continuous stream of real-time data from the monitors 10. The database 20 may store instances of the data received at specific times, according to a predetermined sample frequency.

Alternatively, the database 20 may store the received data which is provided by the monitors 10 at a predetermined sample frequency. The database 20 is configured to store as many past instances as can be accommodated according to the storage size of the database 20. In some embodiments, between 5 and 10 past instances of data will be stored for the forecasting process.

The database 20 is further configured to store additional data required by the processing server 30 to predict an expected power output of the network. The database 20 may store configuration information relating to some or all of the solar power generators 100 in the network such as, for example, the size, orientation, inclination, interconnections, installed capacity or inverter efficiency of a solar power generator 100. This information may be used by the model but is not essential for the

computations. The database 20 may also receive and store data which is generated by the processing server 30 in relation to the forecasting process. With reference to Figure 2, the processing server 30 according to the first embodiment is shown, which comprises a normalisation unit 31 for generating a power attenuation map, a motion estimation unit 32 for calculating a corresponding dynamic flow map for attenuation and a forecasting unit 33 for predicting the expected power output of the network of solar power generators 100.

The normalisation unit 31 is configured to generate a power attenuation map using the measured power output of the network received from the plurality of monitors 10. The measured power output is received by the normalisation unit 31 and is converted to a normalised power output based on a maximum power output for the network. The maximum power output is stored in the database 20 and represents the potential power output of the network of solar power generators 100 under a clear sky. The normalised power output represents the actual power output achieved by the network in comparison with the potential maximum. The difference between the normalised power output at a given time and the maximum power output of the network is the power attenuation of the network at that time. The normalisation unit 31 is configured to generate the power attenuation map for the network of solar power generators 100 based on the normalised power output of the network and the location of the solar power generators 100 of the network.

In some embodiments, the database 20 stores a plurality of power attenuation maps generated by the normalisation unit 31, wherein a power attenuation map is generated for each point in time for which the power output of the network has been stored.

The maximum power output stored in the database 20 is a function of time, which accounts for the position of the sun to provide an expected value for the power output at any given time of day on any given day of the year. The database 20 may therefore store up to 365 daily maximum power output graphs for each PV or each monitor in the network. The maximum power output may depend on a location of the solar power generators 100 in the network and additionally on configuration information such as the size, orientation, interconnections, installed capacity or inverter efficiency of a solar power generator. In some embodiments, the maximum output power is pre-determined and stored in the database 20. Alternatively, the processing server 30 may calculate the maximum output power for a certain time, using the location and stored configuration information for the network of solar power generators 100.

In some embodiments, configuration information relating to a newly added solar power generator may be input by the operator in order to update the maximum power output of the network. Alternatively, the one or more monitors 10 in connection with the newly added solar power generator may collect or detect configuration information from the generator. A maximum power output for the newly added solar power generator can be calculated by the processing server 30 based on the configuration information, in order to update the maximum power output of the network.

In some embodiments, the processing server 30 may calculate a maximum power output for a newly added solar power generator for which configuration information is not available. The measured power output of a newly added solar power generator may be compared with the normalised power output of the network to define a maximum power output for the newly added generator. An initial estimate may be made using interpolation, and improved by iteration over time with additional measurements from the connected monitor.

In some embodiments, an iteration process may be used to calculate the maximum power output for the network of solar power generators 100, in the case that configuration information is not provided. The measured power output of the network may be used to provide an initial estimate for the maximum power output, based on a computational model of the network of solar power generators 100. The model of the network may be refined over time as further measurements are received. The system 1 may be configured such that an initial configuration phase is used to monitor power output over a period of time and calculate an expected maximum value for the power output.

A system 1 according to this embodiment of the invention therefore requires no external data beyond the measure power output in order to calculate a maximum output power and calculate the normalised power output of the network. Accordingly, the power attenuation map for the network can be generated by the normalisation unit 31 using the measured power output as the only exogenous data input. With reference to Figure 3, the measured power output data received from a monitor is represented as a plurality of daily graphs showing the intra-day change in power output. As described above, data collected over a period of time, e.g. three days, is used to calculate an initial estimate for the maximum power output. The maximum power output is represented as a daily graph showing the expected intra-day power output for the solar power generator under a clear sky. The maximum power output calculated for a certain day may be modified based on the date and location of the solar power generator to calculate and store an expected maximum power output for another day.

The measured power output data received from the monitor on a given Day X is compared to the calculated maximum power output to generate a normalised power output for the monitor on Day X. Reductions below the expected power output curve are therefore shown as an attenuation in the normalised power output. The normalised power output graphs calculated for the plurality of monitors in the system may also be represented as a plurality of power attenuation maps which show the spatial distribution of attenuation for a single point in time.

In any embodiment of the invention, the maximum power output of the network may be updated over time to account for a change in the performance of the solar power generators 100.

The motion estimation unit 32 is configured to calculate a dynamic flow map for power attenuation representing, in terms of attenuation from the maximum power output, the change in power output of the network over time. The motion estimation unit 32 calculates the dynamic flow map for power attenuation from a plurality of power attenuation maps generated sequentially by the normalisation unit 31. The sequentially generated power attenuation maps may be stored in the database 20. In some embodiments, the motion estimation unit 32 generates the dynamic flow map by application of a contour detection algorithm and a contour motion algorithm. The contour detection algorithm applied by the motion estimation unit 32 is configured to identify discrete regions of attenuation in the plurality of power attenuation maps. The contour motion algorithm is configured to model the non-rigid dynamics of these regions, to calculate a displacement and distortion for each region of attenuation. Attenuation of the normalised power output can be attributed to the presence of clouds and/or aerosols over the geographic region, which causes a decrease in the solar irradiance reaching the network of solar power generators 100. Accordingly, the contour detection algorithm is configured to identify discrete regions of attenuation which can be attributed to individual clouds or areas of high aerosol concentration and the contour motion algorithm is configured to calculate the overall motion and changing shape of the clouds/aerosols over time.

A well known contour detection algorithm is used to generate a plurality of contours or attenuation isolines for the power attenuation map, which bound regions of the map having the same or similar level of attenuation. A region having a uniform level or a peak in attenuation may therefore be attributed to a cloud or a region of high aerosol concentration. The level of attenuation can indicate the density or altitude of the identified clouds/aerosols. Clouds at different altitudes which cause different levels of attenuation of the power output can therefore be identified and tracked independently, even where the associated regions of attenuation overlap spatially.

In some embodiments, the contour motion algorithm comprises a block matching process applied to a plurality of sequentially generated power attenuation maps. Each power attenuation map is segmented into several small regions (blocks), the size of which is defined according to the spatial resolution of the image. The contour motion algorithm defines the displacement of each block between sequential power attenuation maps and calculates a corresponding velocity for each block. The displacement for a block is defined by comparing the block from a first map to a plurality of blocks having a corresponding size in a subsequent map. A maximum search distance may be defined according to the temporal resolution of the measured power output data, to improve accuracy of the algorithm. A metric for the best match within the allowed search distance can be defined by the minimisation of a cost function. Exemplary metrics include the sum of squared differences, mean squared error, cross correlation coefficient and the absolute value of differences.

In some embodiments, the calculated displacement of a block may be used as an indicator for subsequent blocks of the same map, in order to reduce the search area. Alternatively, or in additional, external factors may be used, such as the prevailing wind direction for the region, to further improve the accuracy of the contour motion algorithm.

The contour motion algorithm further determines the development of the size and shape of each region of attenuation. A detected contour in a power attenuation map can be identified according to the displacement of blocks which constitute the

corresponding contour in an earlier map. The contour motion algorithm compares the contours identified in sequential power attenuation maps in order to define large deformations or splitting of the region of attenuation, as well as the creation of new regions.

The motion estimation unit 32 determines the displacement and deformation of identified contours between sequentially generated power attenuation maps, in order to calculate a dynamic flow map for power attenuation across the network of solar power generators 100. The motion estimation unit 32 is configured to calculate the dynamic flow map based on at least two sequentially generated power attenuation maps. In some embodiments, additional power attenuation maps may be used to improve the accuracy of the calculation process. Any number of power attenuation maps may be used provided the overall computation time of the system 1 remains lower than the time resolution of the measure power output, so that a forecast can be provided by the system 1 in real time. In some embodiments, power attenuation maps from data for the previous 10 minutes is used, for example, up to 10 samples collected at 1 minute intervals or 20 samples collected every 30 seconds may be used.

The forecasting unit 33 is configured to predict an expected power output of the network at a future point in time, based on the dynamic flow map for power attenuation calculated by the motion estimation unit 32.

In the generated power attenuation maps, the attenuation of the normalized power output can be attributed to the presence of clouds and/or aerosols over the geographic region, thus causing a decrease in the solar irradiance reaching the network of solar power generators 100. The forecasting unit 33 therefore predicts the expected size, shape and position of clouds/aerosols at a future time by extrapolating the motion of regions of attenuation calculated by the motion estimation unit 32. Based on the calculated dynamic flow map for power attenuation, the forecasting unit 33 generates a predicted set of flow vectors for a future point in time which can be applied to the most recent power attenuation map to generate an expected power attenuation map for that point in time.

The forecasting unit 33 applies the expected power attenuation map to the maximum power output for the network of solar power generators 100 at the corresponding point in the future, in order to predict the expected power generation for the network of solar power generators 100.

In this way, the present invention provides an accurate forecast for the power output of a network of solar power generators 100 which is based only on the measured power output of the network. The system 1 does not require additional exogenous data to provide an accurate prediction, such as meteorological data, which can be difficult or costly to obtain. An accurate forecast of the power output of the network reduces the need to restrict overall power output to match the expected value, which would reduce network efficiency, or compensate using excess energy production by base load power plants, which would lead to increased C0₂ emissions arising from the excess energy production.

The system 1 according to the present invention is particularly accurate over very short- term time horizons. The time-scale over which the invention is accurate is dependent on the size and density of the network and the rate of change of atmospheric conditions. The invention is therefore useful for intra-day predictions and, in particular, for providing forecasts for a time horizon up to 1 hour. On these timescales, base load power plants are unable to respond to rapid changes in demand, however, the invention provides accurate forecasting, thereby avoiding the need for excess electricity production and reducing C0₂ emissions. With reference to Figure 4, a method according to the second embodiment is shown, for predicting the power output from a network of solar power generators. The method comprises measuring the power output of the network (Si), storing the measured power output of the network (S2) and predicting an expected power output of the network (S10). The method of the present invention provides an accurate forecast for the power output of a network of solar power generators based only on the measured power output of the network. The invention avoids the need for exogenous data, such as meteorological data, and allows an operator of the network to supply reliable amounts of energy based on the predictive method. Thus, the efficiency of a network of solar power generators may be improved by reducing the compromises otherwise required to avoid penalty impositions for inaccuracy.

In an embodiment of the present invention, the method starts by measuring the power output of the network of solar power generators (Si). In an optional configuration phase, the power output of the network of solar power generators is monitored over a period of time, and an expected maximum value for the power output is calculated (S3). An initial estimate for the maximum power output, based on the measured power output and a computational model of the network, is refined iteratively by further measurements over the configuration phase.

After the optional configuration phase, the measured power output of the network is normalised with respect to a stored maximum power output of the network (S4), to calculate a normalised power output of the network of solar power generators. A power attenuation map is generated (S5), based on the difference between the normalised power output and the maximum power output.

Attenuation of the normalised power output is attributed to the presence of clouds and/or aerosols over the geographic region, which causes a decrease in the solar irradiance reaching the network of solar power generators. In an embodiment of the present invention, a contour detection algorithm is applied to a plurality of sequentially generated power attenuation maps, in order to identify regions of attenuation (S6). A contour motion algorithm is applied to determine the displacement and deformation of regions of attenuation between sequential power attenuation maps (S7), and a dynamic flow map for power attenuation is calculated (S8).

Finally, the dynamic flow map is extrapolated in time and applied to the most recent power attenuation map, in order to predict an expected power attenuation map for a future point in time (S9). The expected power attenuation map and a maximum power output for the future point in time are used to predict an expected power output of the network of solar power generators (S10). The method according to the present invention provides an expected power output which is particularly accurate over very short-term time horizons. The invention is therefore useful for intra-day predictions and, in particular, for providing forecasts for a time horizon up to 1 hour. On these timescales, base load power plants are unable to respond to rapid changes in demand, however, the invention provides accurate forecasting, thereby avoiding the need for excess electricity production and reducing C0₂ emissions. The implementation of the present invention is not limited only to the energy sector but it can be used in several alternative sectors, where knowledge of cloudiness and solar irradiance is useful. The organisation of outdoors events such as sporting activities, concerts or banquets could benefit from accurate forecasts of cloudiness for the specific area. For example, a local prediction of cloudiness may be particularly useful to the organisers of a tennis tournament.

Agriculture can take advantage of very short-term forecasts of solar irradiance for specific tasks such as fertiliser spraying, threshing, sun drying and cleaning grain where clear sky conditions are desired. The prediction of cloud cover in an area can also be used as an input to improve the prediction of local temperature. In particular, the predicted solar irradiance can be used to infer a measurement of soil temperature, which is interrelated to solar irradiance.

Furthermore, the knowledge of cloudiness at high spatial resolution is essential for aviation when flying under Visual Flight Rules (low level or non-instrument flights). Likewise, the knowledge of the state of the sky is important for outdoor shootings for movies or photography settings, since it influences the lighting situation and the overall mood of the scene. In particular, it is crucial for aerial photography and shoreline mapping.

The dynamic flow map which is generated by the invention can be used to provide very short-term weather prediction information relating to wind speed and direction. The speed and direction of features in the dynamic flow map can be correlated with cloud types and cloud motion. Thus information relating to wind speed and direction at known altitudes may be extracted. The present invention could also be integrated to ensemble models of low cloud motion for monitoring desert dust storms or volcanic ashes and near surface atmospheric pollution. Finally, the present invention deals with big data analytics that could be assimilated in order to develop novel irradiance maps at very high resolution that could upgrade our knowledge for local variations of meteorological conditions.

In addition to the embodiments of the present invention which have been shown and described, it will be appreciated by those skilled in the art that changes may be made in these embodiments without departing from the invention, the scope of which is defined in the appended claims. Various components of different embodiments may be combined where the principles underlying the embodiments are compatible.

Claims

1. A system for predicting the power output from a network of solar power generators, comprising:

monitoring means for measuring the power output of the network;

a database for storing the measured power output of the network; and a processor arranged to receive the measured power output of the network and predict an expected power output of the network;

wherein the processor comprises:

a normalisation unit for generating a power attenuation map;

a motion estimation unit for calculating a dynamic flow map for power attenuation based on a plurality of sequentially generated power attenuation maps; and a forecasting unit for predicting the expected power output of the network based on the measured power output and the dynamic flow map for power attenuation.

2. The system according to claim 1, wherein the normalisation unit is configured to calculate a normalised power output of the network based on the measured power output and a maximum power output for the network stored in the database, and generate the power attenuation map based on the normalised power output of the network.

3. The system according to claim 2, wherein the processor is configured further to calculate the maximum power output for the network of solar power generators based on a time, the location of solar power generators, and configuration information for the network of solar power generators which is stored in the database, the configuration information including at least one of a size, orientation, inclination, interconnections, installed capacity and inverter efficiency of the solar power generators of the network.

4. The system according to claim 3, wherein the monitoring means is configured further to detect the configuration information for the network of solar power generators and the database is arranged to stores the detected configuration information.

5. The system according to claim 2, wherein the processor is configured further to calculate the maximum power output of the network of solar power generators based on the measured power output of the network over a pre-determined period of time.

6. The system according to any preceding claim, wherein the motion estimation unit is configured to generate the dynamic flow map using a contour detection algorithm for identifying regions of attenuation in the plurality of sequentially generated power attenuation maps and a contour motion algorithm for determining the displacement and deformation of the identified regions of attenuation over time.

7. The system according to any preceding claim, wherein the monitoring means is configured to detect a further solar power generator added to the network of solar power generators.

8. The system according to claim 7, wherein a maximum power output of the further solar power generator is estimated by the processor based on a measured power output of the further solar power generator and a spatial interpolation of the power attenuation map.

9. The system according to any preceding claim, wherein the database and processor are integrated in a terminal device.

10. A method of predicting the power output from a network of solar power generators, comprising:

measuring the power output of the network;

storing the measured power output of the network; and

predicting an expected power output of the network;

wherein the prediction of subsequent power output comprises:

generating a power attenuation map;

calculating a dynamic flow map for power attenuation based on a plurality of sequentially generated power attenuation maps; and

predicting the expected power output of the network based on the measured power output and the dynamic flow map for power attenuation.

11. The method according to claim 10, wherein the generating the power attenuation map comprises calculating a normalised power output of the network based on the measured power output and a maximum power output for the network, and generating the power attenuation map based on the normalised power output of the network.

12. The method according to claim 11, wherein the calculating the normalised power output comprises calculating the maximum power output for the network of solar power generators based on a time, the location of solar power generators, and configuration information for the network of solar power generators, the configuration information including at least one of a size, orientation, inclination, interconnections, installed capacity and inverter efficiency of the solar power generators of the network.

13. The method according to claim 12, further comprising detecting the

configuration information for the network of solar power generators and storing the detected configuration information.

14. The method according to claim 11, wherein the calculating the normalised power output comprises calculating the maximum power output of the network of solar power generators based on the measured power output of the network over a predetermined period of time.

15. The method according to any one of claims 10 to 14, wherein the generating the dynamic flow map further comprises:

identifying regions of attenuation in the plurality of sequentially generated power attenuation maps, and

determining the displacement and deformation of the identified regions of attenuation over time.

16. The method according to any one of claims 10 to 15, further comprising detecting a further solar power generator added to the network of solar power generators.

17. The method according to claim 16, further comprising estimating a maximum power output of the further solar power generator based on a measured power output of the further solar power generator and a spatial interpolation of the power attenuation map.